| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Gary S. Winzelberg, MD, MPH, Division of Geriatric Medicine, University of North Carolina at Chapel Hill School of Medicine, 141 MacNider Bldg, CB 7550, Chapel Hill, NC 27599-7550. E-mail: garywinz{at}med.unc.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Assisted living • Nursing home • Training • Attitudes • Care provider
Ideally, individuals are in the best position to both define and rate their own quality of life. However, cognitive impairment may prevent residents with dementia from rating their quality of life. Currently, more than 50% of the individuals in nursing homes have dementia, as do 23% to 42% of the individuals in residential care/assisted living (RC/AL) facilities (Zimmerman et al., 2003). Because residents with dementia comprise an enlarging population within long-term care facilities, fostering their quality of life merits particular attention.
Lawton (1994) initially developed a conceptual framework of quality of life in dementia that includes the domains of psychological well-being, behavioral competence, care environment, and perceived quality of life. In long-term care facilities, nursing assistants have a critical role in promoting these quality-of-life domains because they deliver the majority of personal care. Because they are in this key position, the perceptions they have about residents are critically important, as their attitudes may well influence the manner in which they provide care. Unfortunately, a recent study of nurses and nurse aides found that the five most prevalent perceptions they held of individuals with dementia were negative: They saw them as being anxious, having little control over their behavior, being unpredictable, being lonely, and being frightened and vulnerable (Brodaty, Draper, & Low, 2003). Complicating this situation, nursing assistants receive minimal resident care training and potentially no dementia-specific education, and they themselves recognize the need for more training; in fact, these workers emphasize that training is important to their ability to provide quality care (Schirm, Albanese, Garland, Gipson, & Blackmon, 2000).
Given their lack of dementia-care training, nursing assistants may rate resident quality of life on the basis of negative biases developed from daily interactions with severely impaired individuals rather than from more balanced observations of residents' remaining capabilities. Thus, nursing assistants may perceive quality of life from their own perspectives instead of from resident-centered observations (Boettcher, Kemeny, DeShon, & Stevens, 2004). Although quality-of-life ratings are ideally based on resident preferences and characteristics, factors related to nursing assistants themselves or the facilities in which they work may become important determinants of how nursing assistants both rate quality of life and make decisions for residents with dementia (Corazzini, McConnell, Rapp, & Anderson, 2004). Our aim in this study is to assess the relationship between resident, nursing assistant, and facility-level characteristics and nursing assistant quality-of-life ratings of long-term care residents with dementia.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Measures
The primary outcome measure was the nursing assistants' rating of resident quality of life, according to the Quality of Life–Alzheimer's Disease scale (QOL-AD; Logsdon, Gibbons, McCurry, & Teri, 2002). This instrument evaluates the following domains: physical condition, mood, interpersonal relationships, ability to participate in meaningful activities, and financial situation. The original instrument contains 13 items and was designed for community-based settings. A modified version more appropriate for long-term care changed the item "marriage relationship" to "relationships with people who work here," "ability to do chores" to "ability to keep busy," and "ability to handle money" to "ability to take care of self." In addition, 2 new items, "ability to live with others" and "ability to make choices in (one's) life," were added (Edelman, Fulton, Kuhn, & Chang, 2005, this issue). The modified QOL-AD includes 15 items, each of which is rated on a 4-point scale (1 = poor to 4 = excellent), with the total score ranging from 15 to 60. The internal consistency reliability of this measure was excellent (
= 0.88) and the interrater reliability was 0.99 (intraclass coefficient, n = 20). Further, the care provider QOL-AD ratings were correlated with their quality-of-life scores by use of other established instruments: Ratings were positively correlated with both the positive affect scale (r =.46) and the activity scale (r =.52) of the Quality of Life in Dementia instrument (Albert et al., 1996), and there was also a strong positive correlation (r =.68) with the Alzheimer Disease-Related Quality of Life measure (Rabins, Kasper, Kleinman, Black, & Patrick, 2000). This and other information related to the QOL-AD is reported in this issue (Sloane et al., 2005).
We evaluated variables at three levels (facility, resident, and nursing assistant) for their relationship to nursing assistants' QOL-AD scores. At the facility level, these variables included facility type (RC/AL facility or nursing home) and profit–nonprofit status. At the resident level, in addition to demographic characteristics, we assessed cognitive status by using the Mini-Mental State Exam (MMSE; Folstein, Folstein, & McHugh, 1975), or the Minimum Data Set Cognition Scale (MDS-COGS; Hartmaier, Sloane, Guess, & Koch, 1994) if an MMSE result was unavailable (n = 48). We assessed functional status by using the Minimum Data Set Activities of Daily Living Scale (MDS-ADL; Morris, Fries, & Morris, 1999), depression by using the Cornell Scale for Depression in Dementia (CSDD; Alexopoulos, Abrams, Young, & Shamoian, 1988), pain intensity with the Philadelphia Geriatric Center Pain Intensity Scale (PGC-PIS; Parmelee, Katz, & Lawton, 1991), and behavioral symptoms with the Cohen-Mansfield Agitation Inventory (CMAI; Cohen-Mansfield, 1986). The MDS-COGS, functional status, depression, pain, and behavioral ratings were all provided by a more senior staff member (such as a registered or licensed practical nurse) who supervised the nursing assistant.
We used three established instruments to evaluate nursing assistants' attitudes toward residents with dementia and their work experiences. The Approaches to Dementia instrument assesses attitudes toward caring for people with dementia, and it includes two subscales (Lintern, Woods, & Phair, 2000). Items in one subscale address a respondent's degree of hope for individuals with dementia, and questions in the other subscale assess the degree to which respondents endorse items related to "person-centered care" as opposed to considering that all residents with dementia have the same strengths and limitations. The Staff Experience Working with Demented Residents instrument contains six subscales that measure respondents' satisfaction with their work environment and experiences caring for residents with dementia (Åström, Nilsson, Norberg, Sandman, & Winblad, 1991). Finally, the Work Stress Inventory includes six subscales that assess work experiences during the past 30 days, including relationships with coworkers and satisfaction with work load and scheduling (Schaefer & Moos, 1993). Another article in this issue provides additional details about these three measures (Zimmerman et al., 2005).
Finally, using a series of questions developed for this study, we assessed nursing assistants' confidence in their training to both identify and help residents in multiple domains of dementia care (depression, behavioral symptoms, pain, eating, drinking, mobility, and activity involvement). For each of these areas, staff was asked to rate how well trained they felt both to identify problems affecting their residents (assessment) and to help with those problems (treatment). We scored responses on a 4-point scale and summed them across all areas to compute a training-assessment score and a training-treatment score. We summed these two indices to create an overall training score.
Statistical Analysis
To estimate bivariate associations between facility, resident, and nursing assistant characteristics and QOL-AD scores, we used Pearson correlation coefficients and means (standard errors) for continuous and categorical measures, respectively. We adjusted the standard errors of the means for resident clustering within nursing assistants and nursing assistants within facilities by using Taylor series expansion methods (Woodruff, 1971). We tested the statistical significance of these associations by using linear mixed models including random effects for facility and nursing assistants nested within facilities.
To examine whether the associations between facility and nursing assistant characteristics and QOL-AD scores were independent of resident characteristics, we estimated partial correlation coefficients and repeated the linear mixed models, adjusting for the resident covariates noted earlier: cognitive status, number of activities of daily living (ADL) disabilities, depressive symptoms, pain severity, and frequency of behavioral symptoms. We selected these covariates because they are likely to influence nursing assistants' perceptions and are often associated with quality-of-life ratings. Indeed, in this study the association of each covariate with the QOL-AD yielded a value of p <.1.
We excluded 70 residents from some analyses because data for at least one resident covariate were missing. Comparing residents with and without all covariate data, we found that there were no statistically significant differences in their mean QOL-AD scores (37.0 vs 38.4, p =.57), mean MMSE scores (8.3 vs 6.7, p = 0.90), gender (82% vs 74% female, p =.39), or whether they resided in a nursing home (36% vs 40%, p = 0.51) or for-profit facility (68% vs 76%, p = 0.31). However, residents with all covariate data available were younger than those with missing covariates (mean age 84.2 vs 86.3 years, p = 0.02).
We used a hierarchical linear model to estimate the extent to which variability in QOL-AD scores is explainable by factors related to the resident, to the nursing assistant, and to the facility. The use of a hierarchical model addresses statistical issues involving correlated multilevel data such as these, in which nursing assistants could be caring (and reporting the quality of life) for multiple residents, and study facilities could employ multiple nursing assistants. Our first step in this hierarchical model was to separate the total variability in QOL-AD scores into between-resident, between-nursing assistant, and between-facility components. We accomplished this by fitting a random intercepts model (no fixed effects) that provided estimates of variance and the standard error for each component. We then created a series of models in which resident, nursing assistant, and facility factors were added sequentially to identify significant effects on QOL-AD scores and to explore how the variance partitioning changes upon adding these factors to the regression. After each step, we assessed the change in between-resident, between-nursing assistant, and between-facility variation. If a variance component diminishes substantially after a set of characteristics is included in the model, we can conclude that such variations are in part the result of these characteristics.
We included individual variables that were associated with QOL-AD scores in the unadjusted analysis at p 
.05, with one exception. Although both frequency of residents' behavioral symptoms of dementia and number of depressive symptoms were significantly associated with nursing assistants' QOL-AD ratings, these two variables were also strongly correlated (r =.56, p <.001). We included only frequency of behavioral symptoms in the final model because agitation and other behaviors are more likely to affect nursing assistant ratings of residents with dementia than less easily detected depressive symptoms. Despite not having significant associations with QOL-AD in the unadjusted analysis, we also included the measures related to nursing assistant age, training, approaches to dementia, and staff experience working with demented residents because of hypothesized clinical relevance.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
Table 4 shows the results of the hierarchical linear modeling to predict QOL-AD scores. The first model (no fixed effects) divides the total variability in QOL-AD scores (64.52 = 48.87 + 9.36 + 6.29) into between-resident, between-nursing assistant, and between-facility components prior to the inclusion of any fixed effects (independent variables). In this model, 76% (48.87/64.52) of the overall variability in QOL-AD scores is due to differences among residents, 14% (9.36/64.52) results from differences among nursing assistants, and 10% (6.29/64.52) is from differences among facilities. Thus, a large amount of variation in QOL-AD scores occurs at the resident level (p <.001), and small amounts of variation occur among nursing assistants (p =.07) and among facilities (p =.07).
|
The next model assessed nursing assistant characteristics adjusted for the resident characteristics in the model. The nursing assistant factors were age of the nursing assistants and their summary scores on the Training, Work Stress Inventory, Approaches to Dementia, and Staff Experience Working With Demented Residents instruments. The nursing assistant variance slightly decreased in comparison with the initial model. Only 3% [(1 – (12.71/13.16)) x 100%] of the variance was explained by the nursing assistant characteristics in the model. None of the nursing assistant characteristics included in the model were statistically significant. The facility variance increased slightly after we added nursing assistant variables from the model that adjusted only for resident characteristics. The estimates of the magnitude of the resident characteristics and the resident-level variance change little when we add nursing assistant and facility characteristics to the model.
The final model evaluated facility factors after adjustment for resident and nursing assistant characteristics. The nursing assistant variance increased slightly from the model that did not adjust for facility factors. The facility variance decreased slightly; 25% [(1 – (2.78/3.70)) x 100%] of the facility-level variance was explained by the facility factors in the model. With respect to the initial model that had no fixed effects, 56% [1 – (2.78/6.28) x 100%] of the variability in QOL-AD scores attributable to between-facility characteristics was explained, whereas virtually none of the between-nursing assistant QOL-AD score variability was explained by the measures included in the analysis. In this final model, only resident characteristics had a significant effect in explaining QOL-AD score variation; neither nursing assistant nor facility characteristics were statistically significant.
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Providing person-centered care has been emphasized as important to the quality of care and life of residents with dementia (Kitwood, 1997). In this practice model, care providers form relationships with residents with dementia and they seek to understand and address their individual needs despite their functional and cognitive deficits (Touhy, 2004). The goals of person-centered care and quality-of-life perception are complementary, as each focuses on resident individuality and attempts to avoid generalizations about residents—generalizations that are often negative in nature.
he positive association between attitudes of nursing assistants regarding person-centered care and their resident QOL-AD scores may indicate that care providers who perceive residents with dementia as having the capacity to engage in relationships and activities also will consider that their lives have quality. These care providers may be more likely to provide person-centered care, although this study did not examine care providers' actual treatment of residents. Alternatively, it is possible that those care providers who value the quality of life of residents with dementia will perceive them as individuals. Therefore, the directionality of this relationship is unclear.
In addition, nursing assistants with greater confidence in their training to assess and treat residents' personal care needs had higher resident QOL-AD ratings. Because training confidence may serve as a proxy for perceptions by care providers of their care quality, nursing assistants may believe that resident quality of life depends on the quality of their care. Further, it is possible that training regarding person-centered care principles will result in improved quality-of-life ratings. Person-centered care training would seek to minimize biases held by nursing assistants about dementia that affect their perceptions of resident quality of life. Thus, training to address and treat care needs such as ADLs should monitor not only the competence of care providers but also their confidence. Once again, because directionality is uncertain, it should be considered that nursing assistants who value resident quality of life feel less overwhelmed by, and better able to address, resident quality-of-life needs. Either way, valuing quality of life appears likely to be beneficial to care provision, although the relationship between quality-of-life perceptions and care quality merits further study.
Although the attitudes of nursing assistants were related to their QOL-AD scores, resident clinical status was the most significant predictor of care provider QOL-AD ratings. As resident functional and cognitive impairments increased, QOL-AD scores progressively decreased. In bivariate analyses, the two characteristics associated with the lowest mean QOL-AD scores were dependence in six to seven ADLs and very severe cognitive impairment. In addition, residents with depression and behavioral symptoms received lower QOL-AD ratings from their care providers.
Prior studies of community-based residents with chronic illness and dementia have demonstrated that functional status and depression are strongly associated with both resident and family caregiver quality-of-life ratings (Logsdon et al., 2002; Patrick, Kinne, Engelberg, & Pearlman, 2000). In one of these studies, cognitive impairment severity was not correlated with QOL-AD scores among community-dwelling individuals (Logsdon et al.). Thus, as measured by the QOL-AD, it is possible that dementia may have a greater perceived impact on quality-of-life ratings in long-term care. That is, nursing assistants may be comparing multiple residents in their care when rating any single resident's quality of life. A family member's rating for a loved one is based on a long-term relationship with that individual, and therefore it may be attenuated by the slow trajectory of decline witnessed over time.
It is also important to note that QOL-AD scores from long-term care staff in this study were unbiased by residents' or nursing assistants' demographic characteristics, including age, gender, or race. However, the racial diversity among the residents was limited (8% were African American), which generally reflects the distribution of residents in RC/AL and nursing homes (Howard et al., 2002). Thus, the relationship of quality of life to residents' race is one that has yet to be sufficiently examined. Nonetheless, the results of this study suggest that nursing assistants focus on residents' clinical characteristics when they are assessing quality of life.
It should be recognized that a low quality-of-life rating may be appropriate and consistent with residents' own values. Low scores can provide an opportunity to reassess residents' care plans and ensure that their treatment decisions are consistent with resident preferences, are focused on areas important to quality of life, and are not unduly burdensome. In addition, decreases in quality-of-life perceptions may prompt a discussion of palliative care approaches with residents' families, and thus serve as an indicator of care needs, rather than as an outcome of care, as is often the case.
Although this study demonstrates significant associations between QOL-AD scores and both resident and care provider characteristics, another important finding is that a substantial portion of the QOL-AD score variability among residents with dementia was not explained by the characteristics included in the hierarchical linear model. Although most of the variability [48.87/64.5 x 100% = 76%; see Table 4] in care provider QOL-AD scores results from differences among residents, only 30% of the variance was explained by covariates in the model. The unexplained QOL-AD variability may result, in part, from one or more unmeasured covariates, or measurement error for the covariates included in this study. Given the importance of understanding the factors associated with nursing assistant quality-of-life observations, future research should further examine the factors associated with variability in QOL-AD ratings. Researchers may identify unmeasured covariates in future studies by including measures of resident behavior other than behavioral symptoms, such as the degree of residents' physical contact and engagement (Bradford Dementia Group, 1997; Sloane et al., 1998). Refining the instruments used to assess care provider attitudes and training confidence may be another approach to reducing the variability in QOL-AD scores.
Research efforts to define and measure quality of life in residents with dementia remain in their early stages (Whitehouse, Patterson, & Sami, 2003). To our knowledge, this study is the first to assess how care providers rate the quality of life of long-term care residents with dementia. In this study we found that, although approximately one fourth of the variability of QOL-AD scores was attributed to variations among care providers or variations among facilities, the overall care provider and facility variation was mostly unexplained by the care provider and facility characteristics examined. In addition to identifying further care provider and resident factors associated with quality-of-life ratings, future studies may examine whether improving care provider person-centered attitudes and training results in improved resident quality-of-life ratings. Given that optimal quality of care is a fundamental goal for residents with dementia, the association between nursing assistant quality-of-life ratings and care quality also should be evaluated.
|
Footnotes |
|---|
1 Division of Geriatric Medicine, University of North Carolina, Chapel Hill.
2 Cecil G. Sheps Center for Health Services Research, University of North Carolina, Chapel Hill.
3 Department of Epidemiology, University of North Carolina, Chapel Hill.
4 Department of Biostatistics, University of North Carolina, Chapel Hill.
5 School of Social Work, University of North Carolina, Chapel Hill.
6 Department of Family Medicine, University of North Carolina, Chapel Hill.
Decision Editor: Richard Schulz, PhD
Received for publication June 28, 2004. Accepted for publication January 3, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. L. Gerritsen, T. P. Ettema, E. Boelens, J. Bos, F. Hoogeveen, J. de Lange, L. Meihuizen, C. J. M. Scholzel-Dorenbos, and R.-M. Droes Quality of Life in Dementia: Do Professional Caregivers Focus on the Significant Domains? American Journal of Alzheimer's Disease and Other Dementias, July 1, 2007; 22(3): 176 - 183. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
